APPENDIX: SOME PROBLEMS IN THE PHILOSOPHY OF SCIENCE
5. CRITICISM OF THE INSTRUMENTALIST VIEW
Berkeley's argument, we have seen, depends upon the adoption of a certain philosophy of language, convincing perhaps at first, but not necessarily true. Moreover, it hinges on the problem of meaning, 25 notorious for its vagueness and hardly offering hope of a solution. The position becomes even more hopeless if we consider some more recent development of Berkeley's arguments, as sketched in the preceding section. I shall try, therefore, to force a clear decision on our problem by a different approach--by way of an analysis of science rather than an analysis of language.
My proposed criticism of the instrumentalist view of scientific theories can be summarized as follows.
Instrumentalism can be formulated as the thesis that scientific theories-the theories of the so-called 'pure' sciences--are nothing but computation rules (or inference rules); of the same character, fundamentally, as the computation rules of the so-called 'applied' sciences. (One might even formulate it as the thesis that 'pure' science is a misnomer, and that all science is 'applied'.)
Now my reply to instrumentalism consists in showing that there are profound differences between 'pure' theories and technological computation rules, and that instrumentalism can give a perfect description of these rules but is quite unable to account for the difference between them and the theories. Thus instrumentalism collapses.
The analysis of the many functional differences between computation rules (for navigation, say) and scientific
theories (such as Newton's) is a very interesting task, but a short list of results must suffice here. The logical relations which may hold between theories and computation rules are not symmetrical; and they are different from those which may hold between various theories, and also from those which may hold between various computation rules. The way in which computation rules are tried out is different from the way in which theories are tested; and the skill which the application of computation rules demands is quite different from that needed for their (theoretical) discussion, and for the (theoretical) determination of the limits of their applicability. These are only a few hints, but they may be enough to indicate the direction and the force of the argument.
I am now going to explain one of these points a little more fully, because it ____________________
25 For this problem see my two books mentioned here in footnote 23, and chs. 1, 11, and 14 of the present volume.
-111-
gives rise to an argument somewhat similar to the one I have used against essentialism. What I wish to discuss is the fact that theories are tested by attempts to refute them (attempts from which we learn a great deal), while there is nothing strictly corresponding to this in the case of technological rules of computation or calculation.
A theory is tested not merely by applying it, or by trying it out, but by applying it to very special cases--cases for which it yields results different from those we should have expected without that theory, or in the light of other theories. In other words we try to select for our tests those crucial cases in which we should expect the theory to fail if it is not true. Such cases are 'crucial' in Bacon's sense; they indicate the cross-roads between two (or more) theories. For to say that without the theory in question we should have expected a different result implies that our expectation was the result of some other (perhaps an older) theory, however dimly we may have been aware of this fact. But while Bacon believed that a crucial experiment may establish or verify a theory, we shall have to say that it can at most refute or falsify a theory. 26 It is an attempt to refute it; and if it does not succeed in refuting the theory in question--if, rather, the theory is successful with its unexpected prediction--then we say that it is corroborated by the experiment. (It is the better corroborated 27 the less expected, or the less probable, the result of the experiment has been.)
Against the view here developed one might be tempted to object (following Duhem 28 ) that in every test it is not only the theory under investigation which is involved, but also the whole system of our theories and assumptions--in fact, more or less the whole of our knowledge--so that we can never be certain which of all these assumptions is refuted.
But this criticism overlooks the fact that if we take each of the two theories (between which the crucial experiment is to decide) together with all this background knowledge, as indeed we must, then we decide between two systems which differ only over the two theories which are at stake. It further overlooks the fact that we do not assert the refutation of the theory as such, but of the theory together with that background knowledge; parts of which, if other crucial experiments can be designed, may indeed one day be rejected as responsible for the failure. (Thus we may even characterize a theory under investigation as that part of a vast system for which we have, if vaguely, an alternative in mind, and for which we try to design crucial tests.)
Now nothing sufficiently similar to such tests exists in the case of instruments or rules of computation. An instrument may break down, to be sure,
____________________
26 Duhem, in his famous criticism of crucial experiments (in his Aim and Structure of Physical Theory), succeeds in showing that crucial experiments can never establish a theory. He fails to show that they cannot refute it.
27 The degree of corroboration will therefore increase with the improbability (or the content) of the corroborating cases. See my "'Degree of Confirmation'", Brit. Jour. Phil. Sci., 5, pp. 143 ff., now among the new appendices of my L.Sc.D., and ch. 10 of the present volume (including the Addenda).
28 See n. 26.
-112-
or it may become outmoded. But it hardly makes sense to say that we submit an instrument to the severest tests we can design in order to reject it if it does not stand up to them: every air frame, for example, can be 'tested to
destruction', but this severe test is undertaken not in order to reject every frame when it is destroyed but to obtain information about the frame (i.e. to test a theory about it), so that it may be used within the limits of its applicability (or safety).
For instrumental purposes of practical application a theory may continue to be used even after its refutation, within the limits of its applicability: an astronomer who believes that Newton's theory has turned out to be false will not hesitate to apply its formalism within the limits of its applicability.
We may sometimes be disappointed to find that the range of applicability of an instrument is smaller than we
expected at first; but this does not make us discard the instrument qua instrument--whether it is a theory or anything else. On the other hand a disappointment of this kind means that we have obtained new information through refuting a theory--that theory which implied that the instrument was applicable over a wider range.
Instruments, even theories in so far as they are instruments, cannot be refuted, as we have seen. The instrumentalist interpretation will therefore be unable to account for real tests, which are attempted refutations, and will not get beyond the assertion that different theories have different ranges of application. But then it cannot possibly account for scientific progress. Instead of saying (as I should) that Newton's theory was falsified by crucial experiments which failed to falsify Einstein's, and that Einstein's theory is therefore better than Newton's, the consistent instrumentalist will have to say, with reference to his 'new' point of view, like Heisenberg: 'It follows that we do not say any longer: Newton's mechanics is false. . . . Rather, we now use the following formulation: Classical mechanics . . . is everywhere exactly "right" where its concepts can be applied.' 29
Since 'right' here means 'applicable', this assertion merely amounts to saying, 'Classical mechanics is applicable where its concepts can be applied'-which is not saying much. But be this as it may, the point is that by neglecting falsification, and stressing application, instrumentalism proves to be as obscurantist a
philosophy as essentialism. For it is only in searching for refutations that science can hope to learn and to advance. It is only in considering how its various theories stand up to tests that it can distinguish between better and worse theories and so find a criterion of progress. (See chapter 10, below.)
Thus a mere instrument for prediction cannot be falsified. What may appear to us at first as its falsification turns out to be no more than a rider cautioning us about its limited applicability. This is why the instrumentalist view may be used ad hoc for rescuing a physical theory which is threatened by
____________________
29 See W. Heisenberg in Dialectica, 2, 1948, p. 333 f. Heisenberg's own instrumentalism is far from consistent, and he has many anti-instrumentalist remarks to his credit. But this article here quoted may be described as an out-and-out attempt to prove that his quantum theory leads of necessity to an instrumentalist philosophy, and thereby to the result that physical theory can never be unified, or even made consistent.
-113-
contradictions, as was done by Bohr (if I am right in my interpretation, given in section ii, of his principle of complementarity). If theories are mere instruments of prediction we need not discard any particular theory even though we believe that no consistent physical interpretation of its formalism exists.
Summing up we may say that instrumentalism is unable to account for the importance to pure science of testing severely even the most remote implications of its theories, since it is unable to account for the pure scientist's interest in truth and falsity. In contrast to the highly critical attitude requisite in the pure scientist, the attitude of
instrumentalism (like that of applied science) is one of complacency at the success of applications. Thus it may well be responsible for the recent stagnation in theoretical physics. (This was written before the refutation of parity.)